AN ITERATIVE ALGORITHM FOR SPLINE INTERPOLATION

One of the fundamental results in spline interpolation theory is the famous Schoenberg-Whitney Theorem, which completely characterizes those distributions of interpolation points which admit unique interpolation by splines. However, until now there exists no iterative algorithm for the explicit computation of the interpolating spline function, and the only practicable method to obtain this function is to solve explicitly the corresponding system of linear equations. In this paper we suggest a method which computes iteratively the coefficients of the interpolating function in its B-spline basis representation; the starting values of our one-step iteration scheme are quotients of two low order determinants in general, and sometimes even just of two real numbers. Furthermore, we present a generalization of Newton's interpolation formula for polynomials to the case of spline interpolation, which corresponds to a result of G. Muhlbach for Haar spaces.